JMIR Publications

ABSTRACT

Background:

The World Health Organisation recommends a package of pregnancy care that includes an obstetric ultrasound scan. However, there are significant barriers to universal access to antenatal ultrasound. These are mainly around the cost and need for maintenance of ultrasound equipment; and a lack of trained personnel. As low cost, hand-held ultrasound devices become more available, the current roadblock is the global shortage of healthcare providers trained in obstetric scanning.

Objective:

Towards improvements in pregnancy and risk assessment for women in underserved regions we are undertaking the Computer-Assisted Low-cost Point-of-care UltraSound (CALOPUS) project, bringing together experts in machine learning and clinical obstetric ultrasound.

Methods:

In this prospective study undertaken in two clinical centres (UK and India), participating pregnant women were scanned and full-length ultrasound was captured. Each woman underwent two consecutive ultrasound scans: the first was a series of simple, standardised ultrasound sweeps (the CALOPUS protocol); immediately followed by a routine, full clinical ultrasound examination that served as the comparator. We describe the development of a simple-to-use clinical protocol designed for non-expert users to assess fetal viability, detect the presence of multiple pregnancy, evaluate placental location, assess amniotic fluid volume, determine fetal presentation and undertake basic fetal biometry. The CALOPUS protocol was designed using the smallest number of steps, minimising redundant information, while maximising diagnostic information. We describe how ultrasound videos and annotations are captured for machine learning.

Results:

As of 22nd November 2020, 3880 scans have been acquired from which 1541751 label annotations have been performed.

Conclusions:

The CALOPUS study is a unique study using obstetric ultrasound videos and annotations from pregnancies dated from 11 weeks and followed up until birth using novel ultrasound and annotation protocols. The data from this study is being used for developing and testing a number of different machine learning algorithms to address key clinical diagnostic questions pertaining to obstetric risk management. We also highlight some of the challenges and potential solutions to interdisciplinary multinational imaging collaborations.